Association of mouse Dlg4 (PSD-95) gene deletion and human DLG4 gene variation with phenotypes relevant to autism spectrum disorders and Williams' syndrome

Abstract

Objective: Research is increasingly linking autism spectrum disorders and other neurodevelopmental disorders to synaptic abnormalities ("synaptopathies"). PSD-95 (postsynaptic density-95, DLG4) orchestrates protein-protein interactions at excitatory synapses and is a major functional bridge interconnecting a neurexinneuroligin-SHANK pathway implicated in autism spectrum disorders.

Method: The authors characterized behavioral, dendritic, and molecular phenotypic abnormalities relevant to autism spectrum disorders in mice with PSD-95 deletion (Dlg4⁻(/)⁻). The data from mice led to the identification of single-nucleotide polymorphisms (SNPs) in human DLG4 and the examination of associations between these variants and neural signatures of Williams' syndrome in a normal population, using functional and structural neuroimaging.

Results: Dlg4⁻(/)⁻ showed increased repetitive behaviors, abnormal communication and social behaviors, impaired motor coordination, and increased stress reactivity and anxiety-related responses. Dlg4⁻(/)⁻ had subtle dysmorphology of amygdala dendritic spines and altered forebrain expression of various synaptic genes, including Cyln2, which regulates cytoskeletal dynamics and is a candidate gene for Williams' syndrome. A signifi-cant association was observed between variations in two human DLG4 SNPs and reduced intraparietal sulcus volume and abnormal cortico-amygdala coupling, both of which characterize Williams' syndrome.

Conclusions: These findings demonstrate that DLG4 gene disruption in mice produces a complex range of behavioral and molecular abnormalities relevant to autism spectrum disorders and Williams' syndrome. The study provides an initial link between human DLG4 gene variation and key neural endophenotypes of Williams' syndrome and perhaps corticoamygdala regulation of emotional and social processes more generally.

FIGURE 1. Repetitive and Social Behaviors in Dlg4^+/+ and Dlg4^−/− Mice^a

^a In panel A, Dlg4^+/+ showed spontaneous alternation in a T-maze, while Dlg4^−/− repeatedly investigated the same arm (#p<0.05 compared with chance in same genotype; *p<0.05 compared with Dlg4^+/+) (N=8–9). Panel B, Dlg4^−/− groomed more in the homecage but not in the novel cage (*p<0.05 compared with Dlg4^+/+) (N=6–7). Panel C, Dlg4^−/− buried fewer marbles in novel and homecage (**p<0.01 compared with Dlg4^+/+) (N=6–8). Panels D, E, and F, Dlg4^−/− were slower to first ultrasonic vocalization to a proestrous female and made fewer ultrasonic vocalizations overall (**p<0.01, *p<0.05 compared with Dlg4^+/+) (N=6–7). Panel G, genotypes did not differ in social behavior in a free dyadic interaction (N=18). Panel H, both genotypes investigated a mouse more than an empty cage, and a novel mouse over a familiar one, but Dlg4^−/− investigated the mouse and then the novel mouse more than did Dlg4^+/+ (##p<0.01 compared with empty/familiar in same genotype, **p<0.01 compared with Dlg4^+/+) (N=17–26). Panel I, genotypes did not differ in transitions between stimuli. Panel J, genotype differences described in panel H were unaffected by 2-methyl-6-phenylethynylpyridine (MPEP) treatment (#p<0.05 compared with empty/familiar in same genotype, *p<0.05 compared with Dlg4^+/+) (N=5–6). VEH=vehicle (control condition). Panel K, both genotypes investigated an object more than an empty cage, and a novel object over a familiar one (#p<0.05 compared with empty cage, ##p<0.01 compared with object 1) (N=10–11). Data are mean values.

FIGURE 2. Motor Functions and Cerebellar Morphology in Dlg4^+/+ and Dlg4^−/− Mice^a

^a In panel A, Dlg4^−/− traveled less far and spent less time in the center (inset) in a novel open field (*p<0.05 compared with Dlg4^+/+) (N=13–22). Panel B, 2-methyl-6-phenylethynyl-pyridine (MPEP) increased open-field activity in both genotypes, normalizing Dlg4^−/− activity to vehicle-treated (VEH; control condition) Dlg4^+/+ levels (N=7–8). Panel C, genotypes did not differ in homecage locomotor activity (N=9–11). Panel D, Dlg4^−/− had lower accelerating rotarod latencies (#p<0.05 compared with trial 1, *p<0.05 compared with Dlg4^+/+) (N=21–26). Panel E, Dlg4^−/− made more foot-slips on a narrow balance beam (**p<0.01 compared with Dlg4^+/+) (N=11–12). Panel F, Dlg4^−/− had lower wire-hang latency (*p<0.05 compared with Dlg4^+/+) (N=13–22). Panels G and H, genotypes did not differ in gross cerebellar morphology. Parasagittal sections immunostained for calbindin (red=Purkinje cells) and neurofilaments (green=Purkinje cell axons, basket cell axons around Purkinje cell somata [white arrow], and molecular layer [ML] interneuron axons [yellow arrow]) and stained with DAPI (blue=granule cell nuclei in internal granule layer [IGL], and sparse basket and stellate cells in ML) (scale bar=100 μm). PCL=Purkinje cell layer. Panel I, structural MRI found no genotype differences in gross brain morphology or volume. Data are mean values.

FIGURE 3. Anxiety- and Stress-Related Phenotypes, Amygdala Spine Morphology, and Forebrain Gene Expression in Dlg4^+/+ and Dlg4^−/− Mice^a

^a In panel A, genotypes did not differ in dark time or light/dark transitions in the light/dark test (N=9–22). Panel B, Dlg4^−/− spent less time in the elevated plus-maze open arms when started facing a closed arm (N=16–26) and more open time when started facing an open arm (*p<0.05 compared with Dlg4^+/+) (N=11–14). Panel C, Dlg4^−/− showed greater stress-induced hyperthermia (**p<0.01 compared with Dlg4^+/+) (N=7–9). Panel D, Dlg4^−/− had greater corticosterone response to restraint stress (##p<0.01 compared with control, *p<0.05 compared with Dlg4^+/+) (N=4–7). Panels E and F, Dlg4^−/− showed larger headwidth of relatively long dendritic spines in basolateral amygdala neurons but normal spine density distribution (**p<0.01 compared with Dlg4^+/+) (scale bar=10 μm) (N=4–6/genotype, 9–10 neurons, 21–37 dendrites). Panel G, Dlg4^−/− differed in forebrain expression of 10 synaptic and plasticity-related genes (N=3). Panel H, Dlg4^−/− had reduced forebrain protein levels of Cyln2 (**p<0.01 compared with Dlg4^+/+) (N=4). Data are mean values.

FIGURE 4. Human Variation in Cortico-Amygdala Connectivity and Gray Matter Volume Observed by Neuroimaging in DLG4 SNPs^a

^a In panels A and B, DLG4 SNP rs390200 G allele carriers showed lesser subgenual anterior cingulate cortex-amygdala connectivity than did homozygous A allele carriers during a threatening face task (**p<0.01, *p<0.05 compared with AA) (N=115). Panels C and D, homozygous DLG4 SNP rs17203281 G allele carriers had lower gray matter volume near the intraparietal sulcus than did A allele carriers (*p<0.05 compared with AA/AG) (N=93). Data are mean values. ROI=region of interest.